Every lead drops into a single send queue. NuMail routes each message out through whichever of your mailboxes is next in line (Gmail API, Microsoft Graph, or raw SMTP) and never sends faster than a real person would. No shared IPs, no blast relays. Your mailboxes, paced like mailboxes.
Leads enroll once. The scheduler picks the next eligible mailbox per lead, respecting that mailbox's provider, throttle, daily cap, timezone and send window, then dispatches through the right transport.
Every lead in an active campaign step becomes one queued job, tagged with the campaign, the sequence position, and the eligible mailbox pool.
OAuth send via users.messages: no SMTP password, full thread headers.
sendMail on the connected M365 mailbox, with native conversation IDs.
Any provider (custom domains, Zoho, Fastmail), sent SMTP, read over IMAP.
The pipeline isn't one global rate limit. Every mailbox carries its own provider, cadence, daily ceiling, timezone and send window, and the scheduler enforces all five before a single message leaves.
One queue, three transports. A campaign can pool a Gmail mailbox, a Microsoft 365 mailbox and an SMTP custom-domain box together; the scheduler dispatches each send through the correct API for whichever mailbox it lands on. Threading, Message-ID and reply detection are handled per provider, so a lead stays in one clean conversation no matter who sent it.
Each mailbox sends on its own clock: its own daily limit, send window and timezone, plus an optional minimum gap between sends, set per mailbox or per campaign with jitter. Turn on slow ramp and a new box starts at 4 sends a day, adding 4 a day until it reaches its limit; SMTP mailboxes connect with it on. There is no global firehose to misconfigure.
Every mailbox has a timezone. Daily limits reset at that mailbox's local midnight (not UTC, not your dashboard's clock), so a London box and a Denver box roll over independently. Quotas, send windows and "sent today" counters all read from the mailbox's own day boundary.
Hard bounces suppress the address instantly: it's added to the workspace suppression list and never contacted again from any campaign. Soft bounces (full mailbox, greylisting, transient 4xx) are counted against the lead, and a third soft bounce suppresses the address like a hard one. Suppression is workspace-wide and survives re-imports.
Set the hours and days each mailbox is allowed to send, typically business hours, Monday through Friday, in the mailbox's local timezone. Weekends are skipped by default; the queue holds anything scheduled outside the window and releases it the moment the window reopens, without bunching. A campaign can also skip major US holidays, so your sending pattern reads human even at a million-a-month volume.
A fresh mailbox blasting hundreds of cold emails on day one is how domains get burned. With slow ramp on, a new box starts at 4 sends a day and adds 4 a day until it reaches its daily limit, then holds there.
Every mailbox connects with a modest daily limit, a business-hours window and the weekend skip already set, and its count resets at its own local midnight. SMTP mailboxes also start on the slow ramp. Change any of it per mailbox when you have the reputation to spend.
Shared-IP relays optimize for throughput and let deliverability be your problem. NuMail optimizes for mailboxes that keep landing in the inbox at scale.
Connect your Gmail, Microsoft 365 and SMTP mailboxes, set a pool, and let the pipeline pace it. Daily limits, send windows and bounce handling are on by default.